[PATCH v2 6/9] mmc: meson-gx: remove unneeded meson_mmc_clk_set in meson_mmc_clk_init

[Michał Zegan]

W dniu 18.02.2017 o 11:37, Heiner Kallweit pisze:
> Am 18.02.2017 um 04:44 schrieb Michał Zegan:
>> Hello Heiner,
>>
>> W dniu 16.02.2017 o 20:42, Heiner Kallweit pisze:
>>> This clk_disable_unprepare isn't right here. At first the condition
>>> should be "if (ret)" as the disable/unprepare is supposed to be
>>> executed if the previous command fails.
>>> And if the previous command fails and meson_mmc_clk_init returns
>>> an error, then probe jumps to label free_host where we have this
>>> disable/unprepare already.
>>>
>>> Signed-off-by: Heiner Kallweit <hkallweit1 at gmail.com>
>>> ---
>>> v2:
>>> - extended commit message
>>> ---
>>>  drivers/mmc/host/meson-gx-mmc.c | 6 +-----
>>>  1 file changed, 1 insertion(+), 5 deletions(-)
>>>
>>> diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
>>> index 68e76fa8..850b0152 100644
>>> --- a/drivers/mmc/host/meson-gx-mmc.c
>>> +++ b/drivers/mmc/host/meson-gx-mmc.c
>>> @@ -320,11 +320,7 @@ static int meson_mmc_clk_init(struct meson_host *host)
>>>  	/* Get the nearest minimum clock to 400KHz */
>>>  	host->mmc->f_min = clk_round_rate(host->cfg_div_clk, 400000);
>>>  
>>> -	ret = meson_mmc_clk_set(host, host->mmc->f_min);
>>> -	if (!ret)
>>> -		clk_disable_unprepare(host->cfg_div_clk);
>>> -
>>> -	return ret;
>>> +	return meson_mmc_clk_set(host, host->mmc->f_min);
>>>  }
>>>  
>>>  static void meson_mmc_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
>>>
>>
>> I see one problem here.
>> As you said, if the previous command (that is the meson_mmc_clk_set
>> function call) fails, then the whole meson_mmc_clk_init returns an error
>> value, and meson_mmc_probe jumps to free_host and disables the clock there.
>> But what happens if the clk_prepare_enable is the line that fails? in
>> this case, the function returns an error value too, and in the free_host
>> meson_mmc_probe tries to disable the clock that is not enabled, and that
>> probably results in a kernel warning because you tried to disable a
>> disabled clock.
>> I was preparing a patch to disable clocks in the probe function only if
>> they were enabled before, but then the meson_mmc_clk_init function
>> remains a problem, because we don't know if the clock was enabled
>> successfully or not when it returns an error.
>>
> Thanks for pointing out, you're right. The solution however should be
> quite easy. I meson_mmc_clk_init we just change the condition of
> the disable/unprepare to "if (ret)" and in probe we remove this
> disable/unprepare from the error path.
> 
If you do that, then if something fails *after* the meson_mmc_clk_init
returns success, but something else fails later, then you will end up
not disabling those clocks. What I thought about is: change the
condition as you said. but in meson_mmc_probe, either check if clock is
enabled in free_host before disabling it, or have two additional error
passes: one used after the core_clk is enabled (currently if it fails
then it will also be a kernel warning), and one used after the
meson_mmc_clk_init returns success that would also disable the div
clock. Not sure if that second one is appropriate or not.
> 
> _______________________________________________
> linux-amlogic mailing list
> linux-amlogic at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-amlogic
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 525 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-amlogic/attachments/20170218/aa2e76bb/attachment.sig>